2. AI Systems

discrete vs. continious

If the environment has a limited number of distinct, clearly defined states then it is called discrete; otherwise it is continuous.

Observable vs. Partially Observable or Unobservable

If it is possible to determine the complete state of the environment at each time point, then it is called observable; otherwise it is only partially observable​ or unobservable.

static vs. dynamic

If the environment does not change while the agents acts -> static

Single Agent vs. Multiple Agents

The environment may contain other agents which may be of the same or different kind as that of the agent.

Accessible vs. Inaccessible

If an agent can obtain complete and accurate information about the state's environment, then such an environment is called an accessible environment else it is called inaccessible.

Deterministic vs. Non-deterministic/Stochastic

If the next state of the environment is completely determined by the current state and the actions of the agent, then the environment is deterministic; otherwise it is non-deterministic.

Episodic vs. Non-episodic/Sequential

In an episodic environment, each episode consists of the agent perceiving and then acting. The quality of its action depends just on the episode itself. In sequential environments the agent requires memory of past actions.

Reflex agent

Selects action on the basis of only the current percept but ignores the percept history.

limited in decision-making, no knowledge about non-perceivables, hard to handle in complex environments

model-based agent

like reflex agent, but better view of environment, track status of environment

not clear how own actions affect world, world model unclear

goal-based agent

build on model-based information, but know desired world (state)

takes decision-making into account

difficult with long sequences of actions to find goal

Utility-based agent

Instead of providing goals, Utility-based agents use a utility function providing a way to rate each action/scenario based on the desired result

help with conflicting goals

• Certain goals can be reached in different ways

• Maps a state or sequence of state onto a real number

Learning Agent

These agents employ an additional learning element to gradually improve and become more knowledgeable over time about an environment.

Components:

1. LearningElement: It is responsible for making improvements by learning from the environment

2. Critic: Gives feedback, describing how well the agent is doing with respect to a fixed mea surement

3. Performance Element: It is responsible for selecting actions

4. ProblemGenerator: Responsible for suggesting actions that will lead to new experiences